search

trefx / 5s-crate

Trusted Workflow Run Crate / specifies a profile of RO-Crate for the purpose of workflow execution in a distributed trusted research environment (TRE).
http://trefx.uk/5s-crate/
MIT License
3 stars 0 forks source link

Output file path not specified #9

Open vpnu opened 1 year ago

vpnu commented 1 year ago

There is no way currently to get the output file path from the incoming crate. It is needed for the Wfexs stage file to be generated.

beforan commented 1 year ago

We also need it to know what data entities to add to metadata / results on the execution CreateAction / hasPart on the root.

e.g. WfExS outputs a crate that contains quite a bit of stuff including but not limited to the outputs defined by the workflow and stage file.

When adding outputs to the main crate, I'm currently proposing we add that entire output in outputs/ but we don't necessarily want to reference all files in the metadata? (I'm not sure we are realistically capable of doing so).

Really, in addition to that folder in general as a Dataset, we should specifically add entities for outputs as defined by the Workflow, and so it would be easiest to get this from a CreateAction outputs definition. These would then be referenced in results and hasPart as per the spec.

stain commented 1 year ago

I don't think all the outputs from WfExS should need to appear again in this crate if that is included as a sub-crate. It is fine to not reference all the files from a crate, you may just describe the folder for instance.

And then "lift" the particular outputs to the Trusted crate's Action - it's consumer do not need to know that those outputs are also in the nested "as-is" crate from WfExS. This is also probably easier as the nested action would have the different ID compared to the original request. (You can link those with sameAs in the top-level crate)

vpnu commented 1 year ago

@stain We are currently trying to get the HutchAgent to write the Wfexs stage file using the incoming Trusted Workflow Run Crate plus the Workflow Run Crate from Workflow Hub once a DownloadAction has occured. The current profile doesn't specify the input and output parameter values that are needed for the stage file to be generated, it only has their names. The values are in the secondary workflow, which is referenced by the main workflow but not explicitly mentioned in either the main incoming crate or the sub-crate.